How to tell unbalance from shaft misalignment: signs, phase and the right order of work

A pump starts humming a month after a repair. The vibration meter on the bearing housing shows 7 mm/s, and someone suggests pulling the impeller off and balancing it. Before you agree, spend twenty minutes on three extra readings. Unbalance and shaft misalignment give similar figures on a single channel, but they are cured by opposite methods: one with weights on the rotor, the other by shifting the motor on the foundation.

Updated 27 August 2026 · by AXILINE · Vila Nova de Gaia

Why these two defects get confused, and why the mistake is expensive

Both defects produce vibration at the running speed. Both grow as speed increases. Both are most often checked the same way: hold a vibration meter against the bearing housing horizontally, get one number in mm/s, and go away to think about it. From a single channel with no phase, the two defects are simply indistinguishable.

That is where they diverge. Unbalance is an uneven distribution of mass around the rotor's axis of rotation: the centre of mass is offset, and rotation produces a centrifugal force that turns along with the rotor. The only way to remove it is with mass: add metal in antiphase, or take it off at the heavy point. Shaft misalignment is a mismatch between the axes of rotation of two shafts joined by a coupling. The force here does not come from mass: every turn, the coupling forces the shafts into a position they do not want to sit in.

A wrong diagnosis costs more than a shift. Balancing a misaligned machine usually makes things worse, and below we go through exactly why. Aligning the shafts of a perfectly balanced machine is pointless, but at least harmless. So if you are unsure of the order, start with alignment, not weights.

The instrument does not make the diagnosis. It gives you numbers: amplitude, phase, the spectrum (the breakdown of vibration by frequency). The diagnosis comes from comparing several points and several operating conditions.

Table: nine signs that tell the two defects apart

No single sign in the table works on its own. Read it as a whole: three or four signs pointing the same way give you a working hypothesis; one on its own gives you nothing.

Axial vibration is the fastest and most underrated sign. One reading along the shaft axis at the bearing housing next to the coupling settles half the question in two minutes.

Direction and phase: three readings that make the diagnosis

The difference between "roughly clear" and "clear" is phase: the angle showing at what point in the turn the vibration reaches its peak. The instrument gives you this alongside the 1x amplitude, and on a two-channel system you get phase at two points at the same time, which is what lets you compare the points against each other. The Balanset-1A measures phase to within about ±1°, so telling 90° from 180° is not a problem.

There is one condition: the sensors have to be mounted identically. The same direction, the same orientation, the same mounting method, the same laser phase sensor, and the same reflective marker on the shaft. Flip a sensor over, and the phase shifts by 180°, handing you a neat but false sign of shaft misalignment.

1. 1

   Horizontal against vertical at one support

   Fit sensors on one bearing housing: one horizontal, one vertical, both as close to the bearing as possible. Take the 1x phase on both. A difference of around 90° is a strong argument for unbalance: the force turns along with the rotor, so the vertical channel lags the horizontal one by roughly a quarter turn. A difference closer to 0° or 180° tells you the vibration is directed along a single line, and that points instead to shaft misalignment, looseness, or a support resonance in one direction.

2. 2

   Axial direction across the coupling

   Move both sensors to the axial direction: one on the motor's bearing housing next to the coupling, the other on the pump's bearing housing next to the coupling. Same orientation, both pointing the same way along the shaft. A phase shift of around 180° means that on every turn, one half of the machine moves forward while the other moves back. That is exactly how a coupling under an angular strain behaves. This is the single strongest sign of shaft misalignment you can get from a portable instrument.

3. 3

   Radial phase on both sides of the coupling

   Both sensors horizontal: one on the motor's bearing housing next to the coupling, the other on the pump's bearing housing next to the coupling. A horizontal difference of around 180° is characteristic of parallel misalignment. Close phases tell you both halves of the machine are vibrating as one unit, which points instead to unbalance or a loosened shared frame.

4. 4

   Check stability before you trust the numbers

   Watch the readings for several tens of seconds. 1x amplitude and phase should not vary by more than 10–15%. If the numbers jump around, the machine is running close to resonance, or there is play somewhere, and no phase conclusion drawn from that data can be trusted. We cover the signs of resonance and how to check it with a coast-down in a separate article.

Different instruments can reference phase in different directions. What matters to you is not the absolute value but the difference between two points, taken with the same instrument in the same run. That is why two-channel acquisition is more convenient here than taking readings one after another.

Types of shaft misalignment: parallel, angular, combined

Shaft misalignment comes in three types, and they look different in vibration terms. Knowing the difference is useful: it tells you which way to shift the motor and which shims to change.

All three share the same causes: imprecise installation, uneven thermal growth as the machine reaches its operating point, a settling foundation, strain from rigidly connected pipework, a soft foot.

Thermal growth means a machine aligned to zero offsets while cold will end up misaligned once at operating temperature. The right approach: align with pre-set cold offsets that compensate for the calculated growth, or check the alignment on the warmed-up machine.

Soft foot: the double of shaft misalignment

A soft foot is a situation where one of the machine's mounting feet does not sit flush against the frame or foundation. Under the foot there is a gap, a tilt, dirt, an old forgotten shim, or a distorted frame. As long as the bolt is loose, everything looks fine. The moment you tighten it, the machine's casing twists.

Here is what makes soft foot so hard to track down. The casing has twisted, so the bearing bores no longer sit on one axis. That means the shaft inside the machine is already skewed, even before you get to the coupling at all. You end up with induced misalignment, distorted fits, a rise in 1x and 2x, and a false picture of looseness. A laser system will show misalignment, you will honestly correct it, and the level will be back within a shift, because the cause is still sitting under the foot.

There is a separate headache: a soft foot will not let you align the machine properly in the first place. You shift the motor, tighten the bolts, and the readings drift, because the casing twists differently every time. That is why you check for soft foot before alignment, not after.

• Loosen the bolts on every foot and inspect the pads: dirt, burrs, paint, old shims, wear marks covering only part of the pad.
• Tighten every foot. Then loosen one bolt at a time in turn and watch how the machine reacts. A dial indicator on the foot will show any lift; a vibration meter will show a change in level and 1x phase.
• If loosening a bolt lifts the foot by more than a few hundredths of a millimetre, that is reason enough to rebuild the shim pack. Follow the machine manufacturer's own requirements.
• Use solid stainless-steel shims, not a stack of a dozen thin foils. A pack of many thin sheets behaves like a spring.
• Check whether the pipework is pulling on the machine. Disconnect a flange and see whether the nozzle moves. Strain from a pipe behaves exactly like a soft foot.
• Check the frame and the foundation: cracks, crumbling concrete, grout that has come away, loose anchors.

Loose fasteners, a soft foot, and cracked feet all produce a spectrum full of harmonics and an unstable, unrepeatable phase. If the 1x phase does not repeat from run to run, there is no point reasoning about unbalance versus shaft misalignment any further: the mechanics come first.

Why balancing a misaligned machine does harm

This is the central idea of the whole article, and it is not about carefulness — it is about vector arithmetic.

The instrument measures a single quantity at the running speed: the overall 1x vector. Every force that acts once per turn adds into it. Real rotor unbalance contributes its own vector. Shaft misalignment also contributes to 1x, with its own phase. The instrument cannot tell them apart, and it never could: to the instrument, it is one arrow on a polar diagram.

You run a trial, the software calculates the influence coefficient (the link between "fitted a weight" and "vibration changed") and gives you a correction weight. That weight is sized to zero out the overall vector. Which means its mass compensates not just for the real unbalance, but for the misalignment's contribution too. The vibration genuinely drops, the software says "within tolerance," and you leave.

Now picture this: a month later, someone properly aligns the machine. The misalignment's contribution to 1x disappears. But the weight that was compensating for it is still on the rotor. And now it acts as a real, physically added unbalance, of the same size and in the opposite phase. Vibration after a correct alignment job ends up higher than it was before any of the work started. The mechanic who did the alignment gets blamed for making things worse.

• The weight compensates for a force that was never the rotor's, so you introduce real unbalance where there was none before.
• The misalignment has not gone anywhere: the coupling, the seals and the drive-end bearing keep wearing under the side load — you have simply removed the indicator.
• The result does not hold. Misalignment shifts with warm-up and load, so the compensation is sized for one thermal state and falls apart in another.
• If the correction was done by drilling, the mistake is irreversible: you cannot weld the metal back on, and on a thin-walled wheel the holes reduce strength on top of everything else.
• You lose the chance to catch the cause early. Shaft misalignment often turns out to be the root cause behind several defects at once, and fixing it clears all of them together.

The reverse situation happens too. If you align the machine and 1x stays high while axial vibration drops to normal, that means real unbalance was hiding under the misalignment the whole time. That is a normal course of events, and balancing is the right call here.

The right order: from the foot to the weights

The order is not arbitrary. Each next step only makes sense once the previous one is closed out — otherwise you end up compensating one thing with another and going in circles.

1. Step 1

   Soft foot, fasteners, pipework strain

   Check that every foot sits flush, the bolt torque, the condition of the frame, the anchors and the grout. Relieve any strain from rigidly connected pipework and ductwork. This is the cheapest step, and it is also the one that most often settles the whole question by itself. As long as the casing twists when you tighten it, neither alignment nor balancing will hold.

2. Step 2

   Shaft alignment accounting for thermal growth

   Align to cold offsets that compensate for the machine's calculated thermal growth, or check the result on the warmed-up machine. While you are at it, inspect the coupling: replace a worn flexible element, damaged pins and bushings, rather than trying to nurse them along until the next scheduled repair.

3. Step 3

   Measurement after alignment, at operating condition

   Warm the machine up, bring it to operating speed, and take a fresh reading: overall vibration — the total level across all frequencies — in mm/s RMS (root mean square), 1x amplitude and phase, the spectrum, and the axial component. This is often where it all ends: the level is already normal, and there is nothing left to balance. Do not skip this measurement — it is exactly what separates necessary work from unnecessary work.

4. Step 4

   Decision: balance or not

   Compare overall vibration with 1x. If 1x accounts for most of the level, and axial has dropped with 2x now negligible, on-site rotor balancing will bring the vibration down. If overall is still much higher than 1x, look at bearings, rubbing, hydraulics or resonance. We cover this criterion and its limits in separate articles on the overall-versus-1x ratio and on the cases where balancing does not help.

5. Step 5

   Balancing in the machine's own bearings

   Choose the number of correction planes from the rotor's geometry, run a trial, check that the response is large enough, fit the correction weights, and run a check. If needed, fine-tune with trim balancing (small final adjustment weights) using the stored influence coefficients.

6. Step 6

   Check measurement and report

   Record the residual 1x, overall vibration against the zones, the speed, the measurement points and directions, and the condition of the supports. Repeat the measurement in two to four weeks: that is exactly what tells a lasting result apart from a temporary one.

One practical tip: if the machine was already balanced before you got to it, without the misalignment being fixed, find and remove any previously fitted weights before you align it. Otherwise, after alignment you will get exactly the raised vibration described above, and you will be hunting for its cause all over again.

Where the signs can mislead: what else produces 2x and axial vibration

Let's be honest about the limits of the method. A noticeable 2x and high axial vibration are not the exclusive signature of shaft misalignment. There are at least six situations that give you the same picture with no misalignment at all, and one more where misalignment is present but 2x is almost absent.

• A bent shaft. 1x dominates, axial is large, axial phases along the shaft diverge in a characteristic way. In the spectrum, this can look like both unbalance and misalignment at once. Dial-indicator run-out checks and a visual inspection are what tell them apart.
• Loose mounting. A comb of harmonics — 2x, 3x and higher — half- and sub-harmonics (components at half the running-speed frequency and below), a pronounced directionality to the vibration, and an unstable phase. Checked by torquing and loosening bolts one at a time.
• A support resonance at twice the running speed. If a natural frequency of the bearing housing or the frame lands close to 2x, a modest 2x balloons by a large factor. Checked with a bump test (striking the stopped machine and recording the response) or a coast-down with amplitude and phase recorded against frequency.
• Vane-pass and blade-pass frequency. On a six-vane extractor fan, the vane-pass frequency produces a peak at 6x, not a misalignment harmonic. Always work out the order and check it against the number of vanes and blades.
• An electromagnetic cause in the motor. A component at twice the supply frequency, 100 Hz on a 50 Hz supply, sits right next to 2x on a two-pole motor. Told apart by the fact that it disappears instantly when the power is cut during coast-down, and by finer spectral resolution.
• A rigid flanged coupling. It can transmit large forces while barely raising 2x, because it has nothing left to flex. Here the diagnosis rests on axial vibration, phase, and a direct alignment measurement.
• A flexible rotor. If the operating speed is close to a critical speed (the speed at which the rotor hits its own resonance and starts to bend), the rotor's shape changes with speed, and signs taken at one speed look different at another. For machines like this, the rules for flexible rotors apply, not the simple criteria for rigid ones.

That gives a simple rule: vibration diagnostics gives you a hypothesis, not a verdict. Shaft misalignment gets confirmed with a direct alignment measurement, not the spectrum alone. Unbalance gets confirmed with a trial run: if the system is linear and the response to the trial weight is predictable, the hypothesis holds.

Sources: ISO 13373-3:2015 · ISO 13373-5:2020 · ISO 20816-1:2016 · ISO 281:2007 · ISO 21940-12:2016

What AXILINE does at your site

You can tell unbalance and shaft misalignment apart yourself: you need a two-channel instrument with phase measurement, half an hour, and discipline in mounting the sensors. This is exactly what we build the Balanset-1A for. The kit includes two accelerometers, a laser phase-and-speed sensor, a two-channel USB module with preamplifiers, integrators and an ADC, and Windows software. The instrument shows overall vibration and 1x separately, amplitude and phase on both channels at once, the time waveform and the FFT spectrum. That set is enough both to tell the defects apart and to do the balancing itself, in one or two planes.

If there is no time to work through this on site yourself, we come to you. Our engineers design and manufacture the Balanset instruments and use them for on-site balancing themselves, so you get the diagnosis and the work from the same people. The order on site is the same as above: measurement and separating the causes first, then the mechanics and shaft alignment, then rotor balancing in its own bearings if needed, and a check run and a report with numbers at the end. If the measurement shows balancing is not needed, we will tell you so.

We give consultation support on questions about the instrument and the method: how to fit sensors to your machine, which direction to choose, how to read the phases you get.

• A reading at the bearing housings: overall vibration in mm/s RMS, 1x amplitude and phase, the spectrum, the axial direction.
• Separating the causes: unbalance, shaft misalignment, soft foot, looseness, resonance, bearings.
• On-site rotor balancing in one or two planes, including splitting the weight across blades and holes and calculating a drilling correction.
• Calculating the tolerance against the G grades (balance quality grades) and assessing overall vibration against the zones, with the applicable part of the standard noted.
• A report with the measurement points, the operating condition, and the initial and residual values.

Which part and edition of the standard applies to your machine needs to be checked separately: the numbers and grades in our materials are working guidance, not a ready-made acceptance standard.

Sources: Balanset-1A manufacturer specification · Balanset-1A operation manual · ISO 20816-1:2016 · ISO 21940-11:2016

Frequently asked questions

Related content